You Ran AI Training. Why Isn't Anyone Using It?

This is the most common thing we hear from organisations that have run company-wide generative-AI training. The room was engaged. People asked questions. And a month later, the people still using it are roughly the people who were already using it.

Blame it on motivation and the conversation ends there. What is actually happening is structural.

Training fixes "didn't know how". That is rarely what stopped them

Training solves a knowledge gap. But the reasons people stall on the floor are usually something else:

• They don't know whether they're allowed to (no rules)
• They don't know where it fits their own work (no translation)
• They don't know whether to trust the output (no verification)
• They have no slack to experiment (no time)

None of those are knowledge problems. They are environment and relationship problems. Two people can sit through the same session and need completely different next steps.

There are patterns to how people stall

RELACTIS maps how people relate to AI as 15 types. Three of them show up constantly when training fails to stick.

The one who says "sounds handy"

They don't object. They are cooperative. They simply never open it — and often cannot articulate why, so months pass. This is usually the largest group in the organisation.

They don't need persuading. They need one concrete five-minute use case handed to them. And if they have genuinely decided not to use it, hearing that reason moves things faster than another session.

The one already using it, ahead of policy

They didn't wait for the company to be ready. The initiative is real, but they may be carrying data and licensing risk personally.

Policing them just pushes the usage out of sight. What works is giving them a sanctioned route and treating their hard-won practice as an organisational asset.

The one criticising without using

The person who keeps saying "AI still isn't good enough" is easy to write off. But the eye for flaws is usually genuine.

Point that eye at verifying output and you have the quality role every organisation is short of. Change the role rather than shutting down the objection.

Look at the distribution before you book the training

One identical session lands as too basic for people already using AI and too abstract for people who are stuck. If you know who is stalling and why, you can change how you deploy it:

• Teams with a translator (the Evangelist) — spread from that person outward
• Teams without one — build a place to ask before you build a curriculum
• Teams with nobody verifying — start by drawing the quality line

Start with your own type

You don't need a company-wide survey to begin. The free assessment takes 24 questions and about four minutes. Run it across a team and you have that team's distribution.

Start by reading the 15 types or taking the free assessment. Rolling it out across an organisation is covered on the business page.